As integrated circuits have become smaller, they have shrunk not only in the amount of surface area required, but also in the thicknesses of the various layers by which they are formed. As the thicknesses of the layers has decreased, it has become increasingly important to planarize a given layer prior to forming a subsequent overlying layer. One of the methods used for such planarization is called chemical mechanical polishing. During chemical mechanical polishing, the surface of the layer to be planarized, thinned, or both is brought into contact with the surface of a polishing pad. The pad and the substrate are rotated and translated relative to each other in the presence of a polishing fluid, which typically contains both physical erosion particles and chemical erosion compounds. Because of the thinness of the layers and the tight tolerances desired, it is important to have a relatively high degree of control over the chemical mechanical polishing process.
One method by which control of the chemical mechanical polishing process is maintained is called conditioning. During conditioning, an implement called a conditioner is brought into contact with the surface of the pad. The conditioner is intended to erode the surface of the pad, so as to expose a portion of the pad that is presumptively more uniform and clean. By this it is meant that the newly exposed portion of the pad is preferably free from surface defects, non-uniformities, and impregnated by-products of the process. Conditioning the pad may be accomplished either between substrate polishing processes, or concurrently with the polishing process. Conditioning tends to generally improve important process characteristics such as substrate to substrate repeatability, polish rate stability, pad life, down time, and overall cost of system ownership.
When a new pad is placed into service, the pad is typically processed with a break in period, during which a more aggressive pad conditioning is performed. Following the break in period, test wafers are processed and checked, and then a standard pad conditioning is accomplished, during which about a tenth of a mil is typically removed from the surface of the pad by the conditioner.
Because the conditioner performs such an important function, it is commensurately important to ensure that the conditioner is functioning properly. Such methods have in the past included a visual inspection of the conditioner, a “fish scale” force monitor, removing the conditioner and performing a flatness test against a known flat standard, and regularly rebuilding or replacing the conditioner. If the conditioner is miss-aligned, worn out, or warped, then it might not make complete and uniform contact with the pad. Such poor pad conditioning might result in poor processing uniformity across a substrate or from substrate to substrate, shorter pad life, increased down time, and other expenses due to yield loss.
Unfortunately, it is very difficult to detect whether the pad conditioner is performing properly, except by the dramatic indicators given above, such as short pad life and wafer non uniformity. Thus, in an extreme condition, a pad conditioner may need to be removed and completely set up anew each day, to ensure that it is in good condition and operating properly. However, this is an expensive and time-consuming process, and opens the door for mistakes to be made during the frequently repeated set up process.
What is needed, therefore, is a system by which proper pad conditioning can be more readily determined.